Multilevel network meta-regression for general likelihoods: synthesis of individual and aggregate data with applications to survival analysis

• DOI: 10.48550/arxiv.2401.12640
• Publication Date: 2024-01-01
• Authors: David Phillippo; Sofia Dias; A. E. Ades; Nicky J. Welton
• Keywords: Meta-regression; Aggregate data

Network meta-analysis combines aggregate data (AgD) from multiple randomised controlled trials, assuming that any effect modifiers are balanced across populations. Individual patient (IPD) meta-regression is the "gold standard" method to relax this assumption, however IPD frequently only available in a subset of studies. Multilevel network (ML-NMR) extends incorporate AgD studies whilst avoiding aggregation bias, but currently requires aggregate-level likelihood have known closed form. Notably, prevents application time-to-event outcomes. We extend ML-NMR individual-level likelihoods form, by integrating function over covariate distributions obtain respective marginal contributions. illustrate with two examples outcomes, showing performance simulated comparison little loss precision full analysis, and demonstrating flexible modelling baseline hazards using cubic M-splines synthetic on newly diagnosed myeloma. general for synthesising individual level networks all sizes. Extension likelihoods, including survival greatly increases applicability method. R Stan code provided, methods implemented multinma package.